Nested Dirichlet process for collaborative mobility modeling

Yiqun Ding,Zhen Zhang,Bin Xu
DOI: https://doi.org/10.1109/ICMLC.2009.5212623
2009-01-01
Abstract:Mobility modeling is the mathematical modeling of mobile users' (cars, cell phone users) movement patterns. The resulting model not only provides us with an understanding of past mobile user movements, but also enables us to predict how a mobile user might move in the future. It has been found useful in both infrastructure-based wireless network and ad hoc network for protocol evaluation, resource planning, etc. A nonparametric hierarchical Bayesian approach is proposed in this paper for extracting hierarchical mobility patterns from mobile user traces. Experiment results show that the proposed method is able to generate a hierarchical mobility model that better reflect the mobility pattern structure in many scenarios. It also has better future movement prediction compared to the hidden Markov model.
What problem does this paper attempt to address?